Research map / Large language models / Agents, code and math
PRM-PBE: Process Reward Model for Reinforcement Learning in Programming-by-Example
Yue Fang, Zhi Jin, Jie An, Hongshen Chen, Jiangmeng Li, Xiaohong Chen, Naijun Zhan
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add external instruction repair in reasoning tree construction (ablation) | PRM-PBE on DeepSeek-Coder-V2 · Lists, MBPP, PROSE average | Pass@1 (%): 44.96 -> 49.79 | Improved |
| add external instruction repair in reasoning tree construction (ablation) | PRM-PBE on Qwen-3 8B · Lists, MBPP, PROSE average | Pass@1 (%): 42.55 -> 47.35 | Improved |
| add preference learning (pairwise ranking) for the PRM (ablation) | PRM-PBE on DeepSeek-Coder-V2 · Lists, MBPP, PROSE average | Pass@1 (%): 47.02 -> 49.79 | Improved |
| add preference learning (pairwise ranking) for the PRM (ablation) | PRM-PBE on Qwen-3 8B · Lists, MBPP, PROSE average | Pass@1 (%): 44.55 -> 47.35 | Improved |
| add three-stage curriculum learning (ablation) | PRM-PBE on DeepSeek-Coder-V2 · Lists, MBPP, PROSE average | Pass@1 (%): 47.55 -> 49.79 | Improved |
| add three-stage curriculum learning (ablation) | PRM-PBE on Qwen-3 8B · Lists, MBPP, PROSE average | Pass@1 (%): 44.94 -> 47.35 | Improved |
| use process reward instead of the best outcome reward model (ORM-Learned) (ablation) | PRM-PBE on DeepSeek-Coder-V2 · five PBE benchmarks | Pass@1 average (%): 49.29 -> 56.61 | Improved |
| use process reward instead of binary outcome reward on D-augmented data (ablation) | PRM-PBE on DeepSeek-Coder-V2 · five PBE benchmarks | Pass@1 average (%): 46.91 -> 56.61 | Improved |
| use RL with PRM instead of SFT | DeepSeek-Coder-V2 · five PBE benchmarks | Pass@1 average (%): 42.76 -> 56.61 | Improved |
| use RL with PRM instead of SFT | Llama-3 · five PBE benchmarks | Pass@1 average (%): 39.30 -> 54.18 | Improved |
| use PRM-PBE instead of GPT-4o WPS prompting | DeepSeek-Coder-V2 · five PBE benchmarks | Pass@1 average (%): 47.49 -> 56.61 | Improved |
| use PRM-PBE instead of SFT on MBPP failure cases | DeepSeek-Coder-V2 · MBPP failures | partial match errors: 33 -> 20 | Improved |
Closest papers by meaning
- Is Programming by Example Solved by LLMs? (NeurIPS 2024 · similarity 0.87)
- P2S: Probabilistic Process Supervision for General-Domain Reasoning Question Answering (AAAI 2026 · similarity 0.87)
- ReasonFlux-PRM: Trajectory-Aware PRMs for Long Chain-of-Thought Reasoning in LLMs (NeurIPS 2025 · similarity 0.87)
- Free Process Rewards without Process Labels (ICML 2025 · similarity 0.87)
- Discriminative Policy Optimization for Token-Level Reward Models (ICML 2025 · similarity 0.86)
- Rewarding Progress: Scaling Automated Process Verifiers for LLM Reasoning (ICLR 2025 · similarity 0.86)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
